Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Optimal timing, cost and sectoral dispatch of emission reductions: abatement cost curves vs. abatement investment Adrien Vogt-Schilb 1 , Guy Meunier 2 , Stéphane Hallegatte 3 1 CIRED, Nogent-sur-Marne, France. ALISS, Ivry-Sur-Seine, France. 3 The World Bank, Sustainable Development Network, Washington D.C., USA 2 INRA–UR1303 Abstract Greenhouse gas emission reductions may be achieved through a combination of actions. Some bring immediate environmental benefits and little consequences over the long term (e.g., driving less), and are best represented by abatement cost curves. Others imply punctual investments and persistent emission reductions (e.g., retrofitting buildings) and are best represented by abatement investment. In a model based on abatement cost curves, efforts to reduce emissions should be equal across sectors, and should be done mostly in the future when the carbon price is higher. In contrast, when reducing emissions requires abatement capital, optimal abatement investment is bell-shaped and concentrated over the short run. Also, the growing carbon price translates in different investment costs (in dollars per ton) across sectors, higher in sectors with larger baseline emissions and sectors where abatement capital is more expensive. Keywords: climate change mitigation; transition to clean capital; path dependence JEL classification: Q54, Q52, Q58 1. Introduction The international community aims at containing global warming below 2◦ C above preindustrial levels. This objective can be reached with different emissions pathways, that is with different distribution of the effort over time. There are also many sectors where greenhouse gas (GHG) emission may be reduced, from the use of renewable energy to better building insulation and more efficient cars. Two important questions for public policy are thus when and where (i.e. in which sectors) emission reductions should occur.1 Many assessments of the optimal timing of greenhouse gas emissions use abatement cost curves (e.g., Nordhaus, 1992; Goulder and Mathai, 2000). In this Email addresses: [email protected] (Adrien Vogt-Schilb), [email protected] (Guy Meunier), [email protected] (Stéphane Hallegatte) 1 A third question, the desirable burden sharing among countries, is not treated in this paper. January 30, 2014 framework, mitigation efforts and the carbon price are almost the same thing, and both grow over time along the optimal pathway. Abatement cost curves embed the implicit assumption that the level of emission abatement may be chosen freely at each time step, without any path dependence (section 2). This assumption is valid for some emission reduction actions, such as consumption pattern changes (e.g., driving less kilometers per year), but not for all actions. Indeed, many authors have re-framed climate mitigation as a transition to a low-carbon economy, where immediate actions have long-term consequences (Grubb et al., 1995; Ha-Duong et al., 1997). Several authors have studied this transition through the lens of knowledge accumulation dynamics (e.g., Gerlagh et al., 2009; Grimaud et al., 2011; Acemoglu et al., 2012). This paper focuses on another dynamic process, the accumulation of abatement capital made to retrofit existing emitting capital (e.g. building insulation, carbon capture and storage) or replace polluting capital (e.g. coal power plants or inefficient cars) by clean capital (windmills and electric vehicles). The analysis builds on a simple intertemporal optimization model where reducing emission requires to invest in and accumulate abatement capital (section 3). Investments in abatement capital have convex costs: accumulating abatement capital faster is more expensive. Also, there is a maximum abatement potential, which is exhausted when all the emitting capital (e.g. energy-intensive dwellings, fossil fuel power plants) is retrofited or replaced by non-emitting capital (e.g. retrofitted dwellings, renewable power). When abatement is obtained through capital accumulation, the optimal carbon price and the optimal abatement efforts have drastically different dynamics, and are not longer growing alongside. The optimal carbon price still grows exponentially over time. The optimal abatement investment, in contrast, is generally bell-shaped: it first grows and then decreases over time. For stringent temperature targets, the bulk of emission-reduction efforts (i.e., investments) happens in the short run. Furthermore, the optimal cost of abatement capital is different from the discounted value of future avoided carbon emissions. Instead, its value is the sum of three terms: (1) the value of avoided emissions; (2) the value of a forgone opportunity, as, given a maximum abatement potential, each investment in abatement capital reduces future investment opportunities; (3) the replacement cost of abatement capital in the long run, which is lower than the carbon price. As a result, determining the optimal level of abatement investments requires more information than just the shadow price of carbon. It also requires the date when all emissions are abated, and the replacement cost of abatement capital in the long term (section 4). The paper also investigates the consequences on the distribution of emission reduction efforts across sectors. The carbon price optimally translates in different investment costs in different economic sectors. On the optimal pathway, short-term efforts, measured in dollars invested per abated ton of carbon, are higher in the sectors that will take longer to decarbonize. These are the sectors with higher abatement potential and sectors where abatement capital is more expensive. A unique carbon price is therefore consistent with higher efforts in sectors with greater baseline emissions, as advocated for instance by Lecocq et al. (1998), Jaccard and Rivers (2007), and Vogt-Schilb and Hallegatte (2014). Finally, numerical simulations based on IPCC data illustrate that the optimal dispatch of emission-reduction efforts (over time and across sectors) is both 2 qualitatively and quantitatively different when assessed using abatement cost functions and abatement investment (section 5). Section 2 provides a brief reminder of the optimal cost and timing of emission reductions in the abatement cost curves framework. Section 3 describes the model of abatement capital accumulation and derive some analytical results concerning the optimal timing of carbon abatement in this new framework. In section 4, we turn to the question of the optimal dispatch of emission reductions across sectors. Section 5 provides numerical illustrations. We conclude in section 6. 2. The abatement cost curve framework A natural idea to study optimal mitigation strategies is to start from an abatement supply curve. For instance, in his seminal contribution, Nordhaus (1992) has framed the question on when and how much to reduce GHG emissions as a cost-benefit analysis, with the Dynamic Integrated model of Climate and the Economy (DICE). In DICE, a social planner chooses at each time step a fraction of GHG emission to abate, and spends a fraction of GDP in emissionreduction activities represented through an abatement cost curve.2 DICE became a reference in the climate mitigation literature, and has been extended in various directions. Among others, Nordhaus (2002) and Popp (2004) investigate the role of induced technical change, and Bruin et al. (2009) study how the explicit description of adaptation in the model changes the optimal timing of emission reductions. Abatement cost curves are found in many other contributions. For instance, Goulder and Mathai (2000) study how induced technical change changes the optimal timing of carbon abatement in an analytical model. Pizer (2002) analyzes the implications of uncertainty surrounding compliance costs on the optimal climate policy path. In this section we use a very simple model based on abatement cost curves, and find that the optimal timing and cost of GHG reductions is essentially the same thing as the exponentially-growing carbon price. 2.1. Abatement cost curve In this framework, the cost of a climate mitigation policy at time t is linked to the abatement at through an abatement cost curve γ. The function γ is classically convex, positive and twice differentiable: ∀at , γ 00 (at ) > 0 (1) 0 γ (at ) > 0 γ(at ) > 0 We make the simplifying assumption that the abatement cost curve γ is constant over time.3 2 This is denoted T C in equation 12 in Nordhaus (1992). Goulder and Mathai (2000) find that the qualitative shape of the optimal timing and cost of emission reductions is robust to this assumption. 3 3 GtCO2/yr Emissions a− a(t) B t time Figure 1: An illustration of the climate constraint. The basic idea behind these curves is that some potentials for emission reductions are cheap (e.g. building insulation pays for itself thanks to subsequent savings), while other are more expensive (e.g. upgrading power plants with carbon capture and storage). If potentials are exploited in the merit order — from the cheapest to the most expensive — the marginal cost of doing so γ 0 (at ) is growing in at , and γ(at ) is convex. 2.2. Abatement potential We explicitly model a maximal abatement potential āt .4 It reflects the idea that GHG emissions cannot be reduced beyond a certain point. For instance, if emissions can be reduced to zero but negative emissions are impossible, āt equals current emissions. ∀t, at ≤ āt (2) 2.3. Carbon budget One approach to determine when and how much to abate carbon emissions is to perform a cost-benefit analysis. Due to the various scientific uncertainties surrounding damages from climate change and climate change itself (Manne and Richels, 1992; Ambrosi et al., 2003), it is frequent to use targets expressed in global warming (as the 2◦ C target from UNFCC), or, similarly (Allen et al., 2009), cumulative emissions (Zickfeld et al., 2009). Here, we constrain cumulative emissions below a given ceiling, a so-called carbon budget B. This keeps the model as simple as possible, and allows us to focus on the dynamics induced by two models of emission reductions (abatement cost curves vs. abatement investment) keeping the dynamics of climate change and climate damages out.5 4 A similar assumption is frequently made implicitly. In particular, in DICE, the abatement cost depends on the fraction of emission abated, and this fraction is caped to 1. 5 Many contributions based on numerical optimisation (including Nordhaus, 1992; Goulder and Mathai, 2000) factor in some climate change dynamics, without changing the qualitative results exposed in this section. 4 For simplicity, we assume that emissions would be constant and equal to ā in absence of abatement.6 We denote mt the cumulative atmospheric emissions at date t. The carbon budget reads (dotted variables represent temporal derivatives): m0 given ṁt = ā − at (3) mt ≤ B B represents the allowable emissions to stabilize global warming to a given temperature target (Matthews and Caldeira, 2008; Meinshausen et al., 2009), but can also be interpreted as a tipping point beyond which the environment is catastrophically damaged. 2.4. The social planner’s program in the abatement cost curve framework In the abatement-cost-curve framework, the social planner determines when to abate in order to minimize abatement costs discounted at a given rate r, under the constraints set by the abatement potential and the carbon budget: Z ∞ min e−rt γ (at ) dt (4) at 0 subject to at ≤ ā (λt ) ṁt = ā − at (µt ) mt ≤ B (φt ) We denoted in parentheses the (positive, present-value) Lagrangian multipliers. 2.5. Results in the abatement cost curve framework With the objective to maintain cumulative emissions below the carbon budget, both the abatement potential ā and the cumulative emission ceiling B are reached at an endogenous date T . ∀t ≥ T, mt = B =⇒ at = ā (from eq. 3) Before this date, the classical result holds: the current carbon price µt ert grows at the discount rate r (Appendix A): ∀t ≤ T, µt = µ (5) This ensures that the present value of the carbon price is constant along the optimal path, such that the social planner is indifferent between one unit of abatement at any two dates. In the abatement cost curve framework, the optimal abatement cost strategy is to implement abatement options such that the marginal abatement cost is 6 Again, Goulder and Mathai (2000) find that the qualitative shape of the optimal cost and timing of emission reductions is robust to this assumption. 5 Figure 2: Optimal timing and costs of abatement in the abatement-cost-curve framework. Left: Before the potential is reached, abatement efforts are equal to the carbon price and grow over time. Right: When the social planner imposes a carbon tax at t0 , the level of abatement “jumps”. equal to the current carbon price at each point in time, until the potential ā and the carbon budget are reached (Appendix A): t ≤ t0 0 (6) γ 0 (a?t ) = µert t0 < t < T 0 γ (ā) t ≥ T Where t0 is the date when the social planner implements the carbon price (Fig. 2). Proposition 1. In the abatement cost curve framework, the optimal abatement pathway is such that: – Both the abatement efforts γ 0 (a?t ) and the abatement level a?t increase over time. – At each time step t, the optimal amount of abatement may be derived from the current carbon price µert and the abatement cost curve γ. – Abatement jumps when the carbon price is implemented. Proof. See (6).7 Such jumps are possible because in the abatement-cost-curve framework, the amount of abatement may be decided at each period independently.8 This simplifying assumption is valid in several cases where abatement action is paid for and delivers emission reduction at the same time, such as driving less or reducing indoor temperatures. In other cases, such as upgrading to more efficient vehicles or retrofitting buildings, costs are mainly paid when the action is undertaken, while annual 7 The result on abatement efforts is general, while the result on the abatement level requires that γ 0 −1 (ex ) is a growing function of x. A sufficient condition is that γ is polynomial. 8 Such jumps in the optimal abatement pathway are present in many works, for instance in the optimal pathway to 1.5◦ C in the original work by Nordhaus (1992) and in the numerical illustrations provided by Goulder and Mathai (2000). Also, Schwoon and Tol (2006) allow explicitly for such jumps, in a model that would otherwise be close to the abatement capital accumulation model presented in section 3. 6 emissions are reduced over several decades. These actions are better modeled as accumulation of abatement capital. In this case, the abatement pathway is continuous (by design, it cannot “jump” anymore); and while the carbon price still grows over time, the cost of the climate policy is bell-shaped (see next section). This yields important consequences on the optimal dispatch of emission-reduction efforts across economic sectors. 3. Abatement investment 3.1. A simple model of abatement capital accumulation In this section, we set up and solve a different model, where all emission reductions require accumulation of abatement capital.9 Note that this is an extreme assumption, useful to compare the results of this model with those from the abatement-cost-curve framework. The stock of abatement capital starts at zero (without loss of generality), and at each time step t, the social planner chooses a positive amount of physical investment xt in abatement capital at , which otherwise depreciates at rate δ: a0 = 0 (7) ȧt = xt − δat (8) For simplicity, abatement capital is directly measured in terms of avoided emissions, such that the carbon budget reads as in section 2 : m0 given ṁt = ā − at (3) mt ≤ B Investment in abatement capital costs c(xt ), where the function c is positive, increasing, differentiable and convex: ∀xt , c00 (xt ) ≥ 0 c0 (xt ) ≥ 0 (9) c(xt ) ≥ 0 The cost convexity bears on the investment flow. This captures increasing opportunity costs to use scarce resources (skilled workers and appropriate capital) to build and deploy abatement capital. For instance, xt will depend on the pace — measured in buildings per year — at which old buildings are being retrofitted at date t (the abatement at would then be proportional to the share of retrofitted buildings in the stock). Retrofitting buildings at a given pace requires to pay a given number of scarce skilled workers. If workers are hired 9 Similar models have been used to study the implications of uncertainty and irreversibility (e.g., Kolstad, 1996; Ulph and Ulph, 1997). Closer to our methods are Fischer et al. (2004) and Williams (2010), who both study the optimal carbon price in the context of abatement investment. Compared to this literature, the contribution of this paper is (i) to disentangle the temporal profile of abatement investment costs from the carbon price, (ii) to study implications concerning the optimal dispatch of emission reductions across sectors, and (iii) to compare the abatement-cost-curve framework with the abatement-investment framework. 7 Figure 3: Optimal cost and timing of abatement in the abatement capital accumulation framework. Left: Abatement efforts are concentrated over the short term. Right: The level of abatement is continuous over time. in the merit order and paid at the marginal productivity, the marginal price of retrofitting buildings c0 (xt ) is a growing function of the pace xt .10 This convexity is of different nature than the convexity of γ in the abatementcost-curve approach presented in section 2, where it arises from heterogeneity in abatement options (e.g., different abatement costs for frequently-driven and occasionally-driven vehicles) while in this model it arises from convex production costs (e.g. for car manufacturers). The social planner chooses when to invest in abatement capital in order to meet a carbon budget at the lowest inter-temporal cost, under the constraint set by the maximum abatement potential (section 2.2): Z ∞ min e−rt c(xt ) dt (10) xt 0 subject to ȧt = xt − δat (νt ) at ≤ ā (λt ) ṁt = ā − at (µt ) mt ≤ B (φt ) Note that the social planner does not control directly abatement at , but investment xt , that is the speed at which emission are reduced. The Greek letters in parentheses are the present-value Lagrangian multipliers (chosen such that they are positive): νt is the value of abatement capital, µt is the cost of carbon emissions, and λt is the cost of the maximum abatement potential. 3.2. Optimal cost and timing of abatement investment Proposition 2. Along the optimal path, abatement increases until it reaches the maximum potential ā at an endogenous date denoted T (Fig. 3). Marginal 10 In this simple model, we do not distinguish between abatement capital that reduces emissions without producing any output, and abatement capital that produces the same output than polluting capital, but without emitting pollution. The cost c(xt ) may thus be interpreted as the cost of investing in abatement capital (e.g., retrofitting an existing building, upgrading a fossil-fuelled power plant with carbon capture and storage), or the cost of building low-carbon capital (e.g., an electric vehicle) instead of, or in replacement for, polluting capital. 8 investment costs depend on this date T , the depreciation rate of the abatement capital δ, and the current carbon price µert : ∀t ≤ T, 0 c (x?t ) rt = µe | Z t ∞ −δ(θ−t) e {z R rt Z dθ − µe | } ∞ T e−δ(θ−t) dθ + e−(r+δ)(T −t) c0 (δā) {z } | {z } O K (11) ∀t ≥ T, x?t = δā Proof. Appendix B Equation 11 states that at each time step t, the social planner should invest in abatement capital up to the pace at which marginal investment costs (Left-hand side term) are equal to marginal benefits (RHS term). The marginal benefits decomposes in three terms. The first term R relates to emission reductions. As before, the current carbon price µert grows at the discount rate. It is multiplied by the quantity R ∞ of GHG saved by the marginal abatement equipment during its lifetime: t e−δ(θ−t) dθ (a naive value of abatement capital). With a higher depreciation rate δ, this benefit is lower, as one investment results in less GHG saved. The second term O comes from a forgone-opportunity effect. The limited potential ā behaves here like a non-renewable resource, an abatement deposit. Each investment in abatement capital brings closer the date T when all emissions are avoided. After T , accumulating more abatement capital does not allow to reduce emissions. The value O of this forgone opportunity is the value of the GHG that the maximum potential prevents to save after T . The shorter it takes to cap all emissions, that is the lower T , the greater is this effect. The third term K is the contribution of the marginal investment to the present value of the final stock of abatement capital (a classical transversality condition). Note that here, the abatement capital is valued at its replacement cost c0 (δā). It is not valued at the carbon price µert because after T , the binding constraint is to maintain abatement capital at its maximum potential, not to reduce emissions. Corollary 1. In general, optimal investment costs do not grow exponentially over time. They may draw a bell shape or decrease over time. Proof. The first order conditions of the problem may be arranged to provide the temporal evolution of marginal investment costs during the optimal transition (B.12) :11 dc0 (xt ) = (r + δ)c0 (xt ) − µert dt (12) Defining t̂, t0 ≤ t̂ ≤ T as the date when investment reaches its maximum, three cases may occur (Fig. 4). In the general case, abatement investment grows until 11 This equation may be called the Euler equation. In section 4.2 we reinterpret it as a condition on the implicit rental cost of abatement capital. 9 ( + ) ′ ̂= >0 < ̂< <0 = ̂ ( + ) ′( ) ̂ Figure 4: The three possible shapes of optimal abatement investment pathways. While the carbon price grows exponentially over time, abatement investment draws a generalized bell shape. it reaches its maximum, when (r + δ)c0 (xt0 ) = µert0 , then decreases to its stead-state value. For stringent climate targets, that is for high carbon prices µ >> (r + δ)c0 (δā), investment starts high and decreases continuously — in Fig. 4, it may be convenient to see (r + δ)c0 (δā) as a parameter of the economy, and µ as a variable dependent on the stringency of the climate target. Only in the asymptotic case of very lax carbon budgets abatement investment grows continuously to its steady-state value. Corollary 1 means that taking into account abatement capital changes drastically the timing of greenhouse gas abatement. It also changes the optimal cost of GHG abatement: Corollary 2. The optimal cost of abatement capital is lower than the value of avoided carbon emissions over its lifetime. Proof. Combining the two first terms (R + O), equation 11 can be rewritten as: ∀t ≤ T, discounted abatement c 0 (x?t ) Z = t T rθ µe |{z} }| { z e−(δ+r)(θ−t) Z ∞ dθ + T carbon price (r + δ) c0 (δā) e−(δ+r)(θ−t) dθ | {z } replacement cost (13) The output of abatement capital (e−(δ+r)(θ−t) ) is valued at the current carbon price (µerθ ) before T , and at the replacement value of abatement capital (r + δ) c0 (δā) after T .12 After T , the replacement cost of abatement capital is lower than the carbon price (Fig. 4) Two important drivers of the cost and timing of abatement investment are the date when the decarbonization is over and the carbon price. The next 12 (r+δ)c0 (δā) can be interpreted as both the rental cost and the levelized cost of abatement capital during the steady state (see next section). 10 section demonstrates that the optimal sectoral dispatch of emission reductions is driven by the different dates when each sector is decabornized. 4. Dispatching emission reductions across sectors In this section, we extend the model of abatement capital accumulation to the case of several sectors. The economy is partitioned in a set of sectors indexed by i. For simplicity, we assume that abatement in each sector does not interact with the others.13 Each sector is described by an abatement potential āi , a depreciation rate δi , and a cost function ci . The social planner’s program becomes: Z ∞ e−rt ci (xi,t ) dt (14) min xi,t 0 subject to ȧi,t = xi,t − δi ai,t (νi,t ) ai,t ≤ āi X (āi − ai,t ) ṁt = (λi,t ) (µt ) i mt ≤ B (φt ) The value of abatement capital νi,t and the cost of the sectoral potentials λi,t now depend on the sector i, while the carbon price µt is still unique for the whole economy. 4.1. Solving for the optimal marginal investment cost The optimal sectoral investment costs are very similar to the optimal cost found in the previous section: Proposition 3. The carbon price grows at the discount rate until all abatement potentials have been reached in each sector i, at the respective and generally different dates Ti . The optimal marginal investment cost is different from the value of avoided emissions: ∀i, ∀t ≤ Ti , 0 ci (x?i,t ) Z = Ti rθ −(δi +r)(θ−t) µe e Z ∞ dθ + t (r + δi ) c0i (δi āi ) e−(δi +r)(θ−t) dθ Ti (15) Proof. Equation 15 is the generalization of equation 13 to the case of several sectors (Appendix C). Corollary 3. Optimal investment costs are higher in sectors that will take longer to decarbonize. 13 This is not entirely realistic, as abatement realized in the power sector may actually reduce the cost to implement abatement in other sectors, using electric-powered capital (Williams et al., 2012). 11 Proof. Seeing (15) as a function, of Ti , one gets: h i dci 0 (x?i,t ) = e−(δi +r)(Ti −t) µerTi − (r + δi )c0i (δi āi ) dTi The term in brackets is positive (see Fig. 4, where the three pathways may now be seen as three different sectors facing the same carbon price). 4.2. Equilibrium decentralization and the principle of equimarginality In the previous subsection, investment costs may differ across sectors, even if there is a unique carbon price. This apparent paradox may be resolved using the following metric: Definition 1. We call marginal implicit rental cost of abatement capital in sector i at a date t the following value: pi,t = (r + δi ) ci 0 (xi,t ) − dci 0 (xi,t ) dt (16) This definition extends the concept of the implicit rental cost of capital (Jorgenson, 1967) to the case where investment costs are endogenous functions of the investment pace.14 It corresponds to the market rental price of abatement capital in a competitive equilibrium, and ensures that there are no profitable trade-offs between: (i) lending at a rate r; and (ii) investing at time t in one unit of capital at cost ci 0 (xi,t ), renting this unit during a small time lapse dt, and reselling 1 − δdt units at the price c0i (xi,t+dt ) at the next time period (see also Appendix D). Proposition 4. In each sector i, before the date Ti , the optimal marginal implicit rental cost of abatement capital equals the current carbon price: ∀i, ∀t ≤ Ti , p?i,t = (r + δi ) ci 0 (x?i,t ) − dci 0 (x?i,t ) = µert dt (17) Proof. Appendix C shows that the first order conditions can be written as: ∀(i, t), (r + δi ) ci 0 (x?i,t ) − dci 0 (x?i,t ) = ert (µt − λi,t ) dt (18) Where λi,t , the Lagrangian multiplier associated with the sectoral potential āi , is null before the potential is exhausted at Ti . Equation 17 may be interpreted as a simple cost-benefit rule. The LHS is the cost of renting the marginal unit of abatement capital during one time period. For instance, it is the premium at which an electricity producer would rent a gas power plant during one year, compared to a coal power plant (and expressed in dollars per avoided ton of carbon). The RHS is the benefit of doing so, that this the price of avoided GHG emissions.15 For instance, it stands for the price 14 We defined marginal rental costs. While market price signals correspond to marginal costs, Jorgenson (1963, p. 143) omits the word ”marginal”. He uses linear investment costs, for which no distinction needs to be done between average and marginal costs. 15 Recall that abatement capital is measured in terms of avoided carbon emissions. 12 of carbon allowances in a well-designed emission trading system. In this sense, the implicit rental cost of abatement capital can be called marginal abatement cost, and (17) simply states that marginal abatement costs should be equal to the carbon price at each point in time and in every sector.16 Prop. 4 also means that the cost-efficiency of investments is more complex to assess when investment costs are endogenous than when they are exogenous. Exposing a case of exogenous investment cost, Jorgenson (1967, p. 145) emphasized: “It is very important to note that the conditions determining the values [of investment in capital] to be chosen by the firm [...] depend only on prices, the rate of interest, and the rate of change of the price of capital goods for the current period.”17 In other words, when investment costs are exogenous, current price signals contain all the information that private agents need to take socially-optimal decisions. In contrast, in the case exposed here — with endogenous investment costs and maximum abatement potentials — the signal given by current prices is incomplete. To determine the optimal amount of investment in a given sector, the carbon price at t must be completed with the correct anticipation of the date Ti when the all emissions in sector i will be capped, and with the longterm replacement cost of abatement capital c0 (δā).18 This remark holds from a private perspective. Take the point of view of the owner of one polluting equipment in a sector i, facing the credibly announced carbon price µert . One question for this owner is when should the equipment be retrofitted or replaced with low-carbon capital: Corollary 4. Along the optimal pathway, individual forward-looking agents in each sector i are indifferent between investing in abatement capital at any time before Ti . Proof. Let τ be the date when the agent invests in abatement capital. Before τ , the agent pays the carbon price. At τ , she invests in one unit of abatement capital at the price c0 (x?τ ). At each time period t after τ , he has to maintain its abatement capital, which costs δc0 (x?t ). The total discounted cost Vi (τ ) of this strategy reads: Z ∞ −rτ 0 ? Vi (τ ) = µτ + e c (xτ ) + e−rt δc0 (x?t )dt (19) τ The first order condition for the individual agent is: Vi0 (τ ) = 0 ⇐⇒ λi,τ = 0 (from eq. 18) (20) This last condition is satisfied when λi,t , the social cost of the sectoral potential āi is null, that is for any τ ≤ Ti (complementary slackness condition C.8). In line with the general theory (e.g., Arrow and Debreu, 1954), Prop. 4 means that the optimal investment pathway is a Nash equilibrium: if forward-looking, 16 The optimal cost of carbon µ itself can be called marginal abatement cost, it measures the discounted cost of tightening the carbon budget by one ton of CO2 . 17 In the present model, these correspond respectively to the current price of carbon µert , the discount rate r, and the endogenous current change of MIC dci 0 (xi,t )/dt. 18 An alternative view is that investment made at t requires to know the full price signal {µerθ − λi,θ }θ∈[t,∞) , as in Eq. 18 (instead of the date Ti as in Eq. 17). 13 cost-minimizing agents correctly anticipate the carbon price µert , optimal investment trajectories xi,t , and the resulting cost of abatement capital ci 0 (x?i,t ), they have no individual interest to diverge from the social optimum. 4.3. An operational metrics: the levelized abatement cost A natural metric to measure and compare the cost of abatement investments in different sectors is the ratio of investment (e.g in dollars) to abated GHG (e.g in tCO2 ). Definition 2. We call Levelized Abatement Cost (LAC) the ratio of marginal investment to discounted abatement. Practitioners often use LACs when comparing and assessing abatement investments, for instance replacing conventional cars with electric vehicles (EV). Assume the additional cost of an EV built at time t, compared to the cost of a conventional car, is 7 000 $/EV. If cars are driven 13 000 km per year and electric cars emit 110 gCO2 /km less than a comparable internal combustion engine vehicle, each EV allows to save 1.43 tCO2 /yr. The the abatement investment cost in this case would be 4 900 $/(tCO2 /yr). If electric cars depreciate at a constant rate such that their average lifetime is 10 years (1/δi = 10 yr) and the discont rate is 5%/yr , then r + δi = 15%/yr and the LAC is 734 $/tCO2 .19 Proposition 5. Levelized Abatement Costs, denoted `t , read: `i,t = (r + δi ) ci 0 (xi,t ) (21) Proof. See Appendix E. Like the implicit rental cost, this metrics can be computed using only current prices (e.g., by myopic investors or myopic regulators). LACs are homogeneous to a carbon price, but unlike the rental cost or the carbon price (see section 4.2), the LACs entirely characterize an investment pathway.20 LACs may be interpreted as MICs annualized using r +δ as the discount rate (taking the carbon price as given, one unit of investment in abatement capital generates a flow of real revenue that decreases at the rate r + δ). Under the assumption that investment costs are linear, LACs computed this way should be equal to the carbon price (see Appendix F). However, generalizing this result to the case of convex investment costs would lead to perverse outcomes. Indeed, LACs should not be equal to the carbon price. Most importantly, a LAC larger than the carbon price may be a clue that more investment, not less, is needed in that sector: Corollary 5. (i) In general, optimal LACs are different in different sectors, and different from the carbon price. (ii) Along the optimal path, if the ratio of investment to abatement (LAC) in a given sector is greater than the current carbon price, then more investment should go to that sector: (r + δi )ci 0 (x?i,t ) =⇒ dci 0 (x?i,t ) >0 dt 19 The investment cost was computed as 7 000 $/(1.43 tCO /yr) = 4 895 $/(tCO /yr); and 2 2 the levelized cost as 0.15 yr−1 · 4 895 $/(tCO2 /yr)= 734 $/tCO2 . `i,t 20 The investment can be calculated from the LAC x 0 −1 . i,t = ci r+δ i 14 Proof. (i) is a consequence of Prop. 3. (ii) see Prop. 5 and Coroll. 1. The following corollary sates that sectors with larger abatement potential and higher investment costs should invest more per abated ton than the others: Corollary 6. The ratio of investment to abatement (levelized abatement cost) should be higher in sectors with larger abatement potential āi or greater marginal abatement investment costs c0i (all other things held constant). Proof. Appendix C.5 demonstrates that these sectors take longer to decarbonize, Coroll. 3 yields the result. In other words, sectors where emissions are more difficult to abate should receive more investment per abated ton of GHG. This may be a relevant qualitative rule of thumb for decision makers willing to schedule or monitor abatement investment. In the next section, we compute numerically optimal investment pathways calibrated on IPCC data. We find that along the optimal pathway, different sectors may invest in abatement capital at drastically different LACs. 5. Illustrative examples using IPCC abatement costs In this section, we use the two models (abatement cost curves as in section 2, or abatement capital accumulation as in section 3) to investigate the optimal sectoral abatements over the 2007-2030 period. We set a policy objective over this period only,21 and use abatement cost information derived from IPCC (2007, Fig. SPM 6). Because of data limitations, this exercise is not supposed to suggest an optimal climate policy. It aims at illustrating the impact of two contrasting approaches to model emission reductions on the optimal abatement strategy — and in particular on the choice of the sectors where short-term emission-reduction efforts should be directed. 5.1. Specification and calibration We extend the problem exposed in section 2 to the case of seven sectors, assuming separate potentials and quadratic abatement costs. Quadratic costs grant that the γi are convex, and simplify the resolution as marginal abatement costs are linear: 1 m 2 γ a 2 i i,t γi 0 (ai,t ) = γim a ∀i, ∀ai,t ∈ [0, āi ] γi (ai,t ) = (22) where γim are parameters specific to each sector. We calibrate these using the abatements corresponding to a 20 $/tCO2 marginal cost in figure SPM.6 in IPCC (2007). We calibrate the sectoral potentials āi as the potential at 100 $/tCO2 provided by the IPCC (this is the higher potential provided for each sector). Numerical values are gathered in Tab. 1. 21 The infinite-horizon models exposed in sections 2 and 3 have to be modified; the results exposed still apply. 15 Abatement potential āi Waste Industry Forestry Agriculture Transport Energy Buildings [ GtCO2 /yr] MAC hparameteri Depreciation rate 34 17.6 15.9 11.9 11.6 10.3 3.6 3.3 4 0.8 5 6.7 2.5 1.7 $/tCO2 GtCO2 /yr γim 0.76 4.08 2.75 4.39 2.1 3.68 5.99 δi [%/yr] MIChparameter i cm i $/tCO2 GtCO2 /yr3 2309 1195 1080 808 788 699 244 Table 1: Numerical values used for the numerical simulations. In the abatement accumulation model, we also assume quadratic costs (and therefore linear marginal costs). 1 m 2 c x 2 i i,t ci 0 (xi,t ) = cm i xi,t ∀i, ∀xi,t ≥ 0, ci (xi,t ) = (23) To calibrate the cm i , we ensure that relative costs (when comparing two sectors) are equal in the two models: ∀(i, j), cm γm i = im m cj γj (24) This defines all the cm i off by a common multiplicative constant. We calibrate this multiplicative constant such that the discounted costs of reaching the same target are equal in the two models (following Grubb et al. (1995)). This way, we aim at reducing differences in optimal strategies to the different models of emission reductions (abatement cost cuves vs. abatement capital accumulation). We call T̄ = 23 yr the time span from the publication date of IPCC (2007) and the time horizon of IPCC data (2030). We set the discount rate to r = 4%/yr. We constrain the cumulative emissions over the period as: Z 0 T̄ X (āi − ai,t ) ≤ B i To compute the carbon budget B, we chose the Representative Concentration Pathway RCP 8.5 (from WRI (20113)) as the emission baseline. An emission scenario consistent with the 2◦ C target is the RCP3-PD.22 We use the difference in cumulative emissions from 2007 to 2030 in this two RCPs to calibrate B = 153 GtCO2 . Finally, we estimate the depreciation rates of capital as the inverse of typical capital lifetimes in the different sectors of the economy (Philibert, 2007; World Bank, 2012, Tab. 6.1). The results are displayed in Tab. 1. 22 Remarkably, the difference in carbon emissions in 2030 between this two RCPs amounts P to 24 GtCO2 /yr, which matches i āi as calibrated from IPCC (Tab. 1). 16 We solve the two models numerically in continuous time. Data and source code will be available at the corresponding author’s web page. Computation and plots use Scilab (Scilab Consortium, 2011). 5.2. Results Fig. 5 compares the optimal mitigation strategy by the two models and Fig. 6 compares the aggregated pathways in terms of abatement and financial effort. Again, these are not optimal pathways to mitigate climate change: they only consider a target in the 2007-2030 window and are based on crude available data. The two models give the same result in the long run: abatement in each sector eventually reaches its maximum potential (Fig. 5). By construction, they also achieve the aggregated abatement target at the same discounted cost. And in both models, the carbon price grows exponentially (Fig. 5, upper panels). However, the strategies from the two models differ radically in terms of the temporal and sectoral distribution of aggregated abatement and costs. The optimal abatement pathway in the abatement-cost-curve framework includes significant abatement as soon as the climate policy is implemented (to emphasize this, we plotted a null abatement between 2005 and the start of the climate policy in 2007). In contrast, the abatement pathway according to the abatement accumulation model starts at zero and increases continuously (Fig. 5, lower panels). In terms of the temporal distribution of abatement costs, the abatement capital accumulation model provides an optimal pathway that starts higher and then decreases over time (Fig. 6, right). Abatement efforts are concentrated on the short term, because once all the emissions in a sector has been avoided using abatement capital, the only cost is that of maintaining the stock of abatement capital. In the abatement-cost-curve framework, the carbon price gives a straightforward indication on where and when efforts should be concentrated. In contrast, the growing carbon price is a poor indicator of the optimal cost and timing of abatement capital accumulation (Fig. 5, higher panels). To avoid paying too much at any point in time, investment should be spread over a larger period, and longer-to-decarbonize sectors should abate at a higher cost than the others. In this example with quadratic investment costs, i.e. with the same cost convexity accross sectors, industry is decarbonized faster (in terms of δi xi,t ) and at a higher cost (`?i,t ) than forestry, despite forestry being a priori cheaper to m decarbonize (in terms of both cm i or (r + δi ) ci , see Tab. 1). This is because industry takes longer to decarbonize, a consequence of its greater abatement potential (Tab. 1). 6. Conclusion Abatement cost curves are appropriate to represent actions with immediate environmental benefits and little consequences over the long-term, such as driving less miles per year, using gas power plants more hours per year and coal power less hours per year, or reducing indoor temperatures when heating is needed. For these actions, the growing carbon price provides a straightforward indication on where and when should be allocated efforts to reduce emissions. Other abatement actions imply investments with long-term consequences, such as replacing thermal cars with plug-in hybrid or electric vehicles, replacing 17 Abatement cost curves Abatement capital accumulation Figure 5: Comparison of optimal abatement strategies to achieve the same amount of abatement, when the costs from IPCC (2007, SPM6) are understood in an abatementcost-curve framework (left) vs. a abatement capital accumulation framework (right). Note: For clarity, we plotted investment in committed MtCO2 /yr (δi xi,t ) instead of crude investment (xi,t in MtCO2 /yr2 ). With this metrics, one thousand electric vehicle built in 2010 that will each save 14 tCO2 during their lifetime count as an investment of 1400 tCO2 /yr in 2010. In the abatement-cost-curve framework, there is no equivalent to the physical abatement investments xi,t , as the planner controls directly the abatement level ai,t . 18 Figure 6: Comparison of the optimal timing and cost of GHG emissions in the two models (abatement cost curves vs abatement capital accumulation). When abatement may be freely chosen on a cost curve at each time step, the abatement can jump from 0 to a large amount instantaneously at the beginning of the period. When abatement requires accumulating capital, abatement has to grow continuously, and the short-term cost are higher than in the abatement-cost-curve framework. We calibrated the two models such that the total discounted cost and the cumulative abatements are equal in the two cases. fossil-fueled power plants with renewable power, or retrofitting buildings. These provide control on the rate of emission reductions, rather than on the emission level directly, and are best modeled as an accumulation of abatement capital. For these actions, the growing carbon price is a more distant indicator of the optimal dispatch of emission reductions over time and across sectors. Indeed, optimal investment is concentrated over the short run, even though the carbon price grows over time. Also, higher short-term efforts (in dollars invested per abated ton) are needed in sectors that will take longer to decarbonize, even though the same carbon price applies to all sectors. Longer-to-decarbonize sectors are those with larger abatement potentials and those where abatement capital is more expensive. Concerning the assessment of abatement investment, the ratio of investment to abatement (that we called levelized abatement cost) should not be used to schedule or monitor abatement investment. Indeed, this ratio is not equal to the carbon price, and not be equal across sector, along the optimal path. Also, the ratio gives a counter intuitive signal: along the optimal path, levelized abatement cost higher than the carbon price mean that more abatement investment should be done. A correct assessment should compare the carbon price to the implicit rental cost of abatement, and requires estimating how long it will take for sectoral emissions to be reduced to their long-term, minimum level. The optimal abatement investment profiles can be interpreted as transitions to clean capital. In the short term, low-carbon capital has to be built to replace (or retrofit) preexisting polluting capital, leading to high investments needs and large transition costs. Over the long term, when the entire stock of emitting capital has been replaced or retrofitted, abatement investments need only to maintain (or grow at the economic growth rate) the stock of clean capital, and the cost is lower than during the transition phase. Our results should be interpreted cautiously, as the model disregards mechanisms that would affect significantly the cost and timing of climate policies. For 19 instance, we did not take into account knowledge accumulation, knowledge spillovers and economic growth. We also disregarded the effect of uncertainty on climate impacts and future technologies, limited foresight by investors, policymakers and regulators, and the limited ability of the government to commit. Finally, we studied abatement cost curve and abatement capital in separate models (for analytical tractability). As already mentioned, a realistic climate strategy will include some actions that are best represented by the former and some by the latter. Further research could thus merge the two approaches.23 Notwithstanding these limitations, this analysis may help clarify public economic issues related to the optimal response to environmental issues. Acknowledgments We thank Alain Ayong Le Kama, Patrice Dumas, Marianne Fay, Michael Hanemann, Frédéric Ghersi, Louis-Gaëtan Giraudet, Christophe de Gouvello, Christian Gollier, Céline Guivarch, Oskar Lecuyer, Baptiste Perissin Fabert, Antonin Pottier, Jean Charles Hourcade, Lionel Ragot, Julie Rozenberg, François Salanié, Ankur Shah, seminar and conference participants at Cired, École Polytechnique, Université Paris Ouest, Paris Sorbonne, Toulouse School of Economics, Princeton University’s Woodrow Wilson School, the Joint World Bank – International Monetary Fund Seminar on Environment and Energy Topics, the Conference on Pricing Climate Risk held at Center for Environmental Economics and Sustainability Policy of Arizona State University, and audiences at the 2013 EAERE and the 2014 ASSA conferences for useful comments and suggestions on various versions of this paper. The remaining errors are the authors’ responsibility. We are grateful to Patrice Dumas for technical support. We acknowledge financial support from Institut pour la Mobilité Durable (Renault and ParisTech), and the ESMAP (The World Bank). The views expressed in this paper are the sole responsibility of the authors. They do not necessarily reflect the views of the World Bank, its executive directors, or the countries they represent. References Acemoglu, D., Aghion, P., Bursztyn, L., Hemous, D., 2012. The environment and directed technical change. American Economic Review 102 (1), 131–166. Allen, M. R., Frame, D. J., Huntingford, C., Jones, C. D., Lowe, J. A., Meinshausen, M., Meinshausen, N., 2009. Warming caused by cumulative carbon emissions towards the trillionth tonne. Nature 458 (7242), 1163–1166. Ambrosi, P., Hourcade, J.-C., Hallegatte, S., Lecocq, F., Dumas, P., Ha Duong, M., 2003. Optimal control models and elicitation of attitudes towards climate damages. Environmental Modeling and Assessment 8 (3), 133–147. Arrow, K. J., Debreu, G., 1954. Existence of an equilibrium for a competitive economy. Econometrica 22 (3), 265–290. 23 In a work in progress, Rozenberg et al. (2013) model both polluting and clean capital accumulation in a general equilibrium framework with economic growth. This allows representing both types of emission-reduction options. 20 Brock, W. A., Taylor, M. S., 2010. The green solow model. Journal of Economic Growth 15 (2), 127–153. Bruin, K. C. d., Dellink, R. B., Tol, R. S. J., 2009. AD-DICE: an implementation of adaptation in the DICE model. Climatic Change 95 (1-2), 63–81. Fischer, C., Withagen, C., Toman, M., 2004. Optimal investment in clean production capacity. Environmental and Resource Economics 28 (3), 325–345. Gerlagh, R., Kverndokk, S., Rosendahl, K. E., 2009. Optimal timing of climate change policy: Interaction between carbon taxes and innovation externalities. Environmental and Resource Economics 43 (3), 369–390. Goulder, L. H., Mathai, K., 2000. Optimal CO2 abatement in the presence of induced technological change. Journal of Environmental Economics and Management 39 (1), 1–38. Grimaud, A., Lafforgue, G., Magné, B., 2011. Climate change mitigation options and directed technical change: A decentralized equilibrium analysis. Resource and Energy Economics 33 (4), 938–962. Grubb, M., Chapuis, T., Ha-Duong, M., 1995. The economics of changing course : Implications of adaptability and inertia for optimal climate policy. Energy Policy 23 (4-5), 417–431. Ha-Duong, M., Grubb, M. J., Hourcade, J. C., 1997. Influence of socioeconomic inertia and uncertainty on optimal CO 2-emission abatement. Nature 390, 271. IPCC, 2007. Summary for policymakers. In: Climate change 2007: Mitigation. Contribution of working group III to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge, UK and New York, USA. Jaccard, M., Rivers, N., 2007. Heterogeneous capital stocks and the optimal timing for CO2 abatement. Resource and Energy Economics 29 (1), 1–16. Jorgenson, D., 1967. The theory of investment behavior. In: Determinants of investment behavior. NBER. Kolstad, C. D., 1996. Fundamental irreversibilities in stock externalities. Journal of Public Economics 60 (2), 221–233. Lecocq, F., Hourcade, J., Ha-Duong, M., 1998. Decision making under uncertainty and inertia constraints: sectoral implications of the when flexibility. Energy Economics 20 (5-6), 539–555. Manne, A. S., Richels, R. G., 1992. Buying greenhouse insurance: the economic costs of carbon dioxide emission limits. MIT Press, Cambridge, Mass. Matthews, H. D., Caldeira, K., 2008. Stabilizing climate requires near-zero emissions. Geophysical Research Letters 35 (4). Meinshausen, M., Meinshausen, N., Hare, W., Raper, S. C. B., Frieler, K., Knutti, R., Frame, D. J., Allen, M. R., Apr. 2009. Greenhouse-gas emission targets for limiting global warming to 2◦ C. Nature 458 (7242), 1158–1162. Nordhaus, W. D., 1992. An optimal transition path for controlling greenhouse gases. Science 258 (5086), 1315–1319. 21 Nordhaus, W. D., 2002. Modeling induced innovation in climate-change policy. In: Technological change and the environment. Grübler, A., Nakićenović, N., Nordhaus, pp. 182–209. Philibert, C., 2007. Technology penetration and capital stock turnover: Lessons from IEA scenario analysis. OECD Working Paper. Pizer, W. A., 2002. Combining price and quantity controls to mitigate global climate change. Journal of Public Economics 85 (3), 409–434. Popp, D., 2004. ENTICE: endogenous technological change in the DICE model of global warming. Journal of Environmental Economics and Management 48 (1), 742– 768. Rozenberg, J., Vogt-Schilb, A., Hallegatte, S., 2013. How capital-based instruments facilitate the transition toward a low-carbon economy : a tradeoff between optimality and acceptability. World Bank Policy Research Working Paper (6609). Schwoon, M., Tol, R. S., 2006. Optimal CO2-abatement with socio-economic inertia and induced technological change. The Energy Journal 27 (4). Scilab Consortium, 2011. Scilab: The free software for numerical computation. Scilab Consortium, Paris, France. Ulph, A., Ulph, D., 1997. Global warming, irreversibility and learning. The Economic Journal 107 (442), 636–650. Vogt-Schilb, A., Hallegatte, S., 2014. Marginal abatement cost curves and the optimal timing of mitigation measures. Energy Policy 66, 645–653. Williams, J. H., DeBenedictis, A., Ghanadan, R., Mahone, A., Moore, J., Morrow, W. R., Price, S., Torn, M. S., 2012. The technology path to deep greenhouse gas emissions cuts by 2050: The pivotal role of electricity. Science 335 (6064), 53–59. Williams, R., 2010. Setting the initial time-profile of climate policy: The economics of environmental policy phase-ins. Working Paper 16120, National Bureau of Economic Research. World Bank, 2012. Inclusive green growth : the pathway to sustainable development. World Bank, Washington, D.C. WRI, 20113. Climate analysis indicators tool (CAIT) version 8.0. World Resources Institute (13). Zickfeld, K., Eby, M., Matthews, H. D., Weaver, A. J., 2009. Setting cumulative emissions targets to reduce the risk of dangerous climate change. Proceedings of the National Academy of Sciences 106 (38), 16129–16134. Appendix A. Proof of Prop. 1 (Abatement cost curves) The Lagrangian associated with (4) reads: Z ∞ Z ∞ X −rt L(at , mt , λt , µt ) = e γ(ai,t ) dt + λt (at − ā) dt 0 0 Z ∞ Z ∞ − µ̇t mt dt + µt (ā − at ) dt + φt (mt − B) 0 0 22 (A.1) The first order condition is: ∂L = 0 ⇐⇒ e−rt γ 0 (at ) + λt − µt = 0 ∂at ⇐⇒ γ 0 (at ) = ert (µt − λt ) ∂L = 0 ⇐⇒ µ̇t = φt ∂mt (A.2) (A.3) The steady state is reached at a date T when ṁt = 0, that is when the abatement potential ā is reached, such that: ∀t < T, at < ā and mt < B ∀t ≥ T, at = ā and mt = B As the associated Lagrangian multiplier, φt is null before the carbon budget is reached (complementary slackness condition): ∀t, =⇒ ∀t < T, φt · (mt − B) = 0 φt = 0 (A.4) This means that the present value of carbon µt is constant while the carbon budget has not been reached (A.3): ∀t < T, µt = µ (A.5) For the same reason, λt is null before the sectoral potential becomes binding: ∀t, =⇒ ∀t < T, Combining (A.6) and (A.2), one gets: ( γ 0 (at ) = λt · (at − ā) = 0 λt = 0 µert t < T γ 0 (ā) t ≥ T (A.6) (A.7) Appendix B. Optimal accumulation of abatement capital (Proof of Prop. 2) Appendix B.1. Lagrangian The Lagrangian associated with (10) reads: Z ∞ Z ∞ L(xt , at , mt , λt , νt , µt ) = e−rt c(xt ) dt + λt (at − ā) dt 0 Z ∞ Z0 ∞ − µ̇t mt dt + µt (ā − at ) dt + φt (mt − B) 0 Z0 ∞ Z ∞ − ν̇t at dt + νt (δat − xt ) dt 0 0 (B.1) 23 Appendix B.2. First order conditions The first order conditions read: c0 (xt ) = ert νt (∂xt ) (B.2) ν̇t − δνt = λt − µt (∂at ) (B.3) µ̇t = φt (∂mt ) (B.4) Where νt is the present value of investment in low carbon capital, µ is the present cost of carbon, and λt is the social cost of the maximum abatement potential. The first order conditions can be rearranged: dc0 (xt ) = ert (ν̇t + r νt ) dt = ert (δνt + λt − µt + r νt ) ert (µt − λt ) = (r + δ) c0 (xt ) − (from B.2) (B.5) (from B.3) (B.6) (from B.2) (B.7) 0 dc (xt ) dt Appendix B.3. Steady state The steady state is reached at a date T when the carbon budget is reached, emission become null, and investment is used to counterbalance depreciation: ∀t ≥ T, ṁt = 0 =⇒ at = ā =⇒ xt = δā (B.8) Appendix B.4. Slackness conditions The complementary slackness conditions mean that the carbon price is constant before the steady state: ∀t < T, mt < B & φt = 0 =⇒ µt = µ (B.9) (B.10) and the social cost of the maximum potential is null before the steady state: ∀t < T, at < ā & λt = 0 (B.11) ∀t ≥ T, at = ā & λt ≥ 0 Appendix B.5. Marginal Implicit Rental Cost of Capital Before T , (B.7) simplifies: ∀t ≤ T, (r + δ) c0 (xt ) − dc0 (xt ) = µert dt (B.12) The textbook solution of this first order linear differential equation is: ∀t ≤ T, c0 (xt ) = e(r+δ)t Z T e−(r+δ)θ µerθ dθ + e(r+δ)t C (B.13) t Where C is a constant. Any C chosen such that c0 (xt ) remains positive defines an investment pathway that satisfies the first order conditions. The optimal investment pathways also satisfies the following boundary conditions. 24 Appendix B.6. Boundary conditions After T , at is constant and the investment xt is used to counterbalance the depreciation of abatement capital. c0 (xT ) = c0 (δā) (from eq. B.8) (B.14) Optimal marginal investment costs (MICs) Injecting B.14 in B.13 and re-arranging, one gets: c 0 (x?t ) rt Z = µe T e−δ(θ−t) dθ + e−(δ+r)(T −t) c0 (δā) (B.15) t Appendix C. Optimal investment in multiple sectors Appendix C.1. Lagrangian The Lagrangian associated with (14) reads: Z ∞ Z ∞ L(xi,t , ai,t , mt , λi,t , νi,t , µt ) = e−rt c(xi,t ) dt + λi,t (ai,t − āi ) dt 0 0 Z ∞ Z ∞ X − µ̇t mt dt + µt (āi − ai,t ) dt + φt (mt − B) 0 Z 0 ∞ − i Z νi,t (δi ai,t − xi,t ) dt ν̇i,t ai,t dt + 0 ∞ 0 (C.1) Appendix C.2. First order conditions The first order conditions read: ci 0 (xi,t ) = ert νi,t (∂xi,t ) (C.2) ν̇i,t − δi νi,t = λi,t − µt (∂ai,t ) (C.3) (∂mt ) (C.4) µ̇t = φt Where νi,t is the present value of investment in low carbon capital, µ is the present cost of carbon, and λi,t is the social cost of the sectoral potential. Appendix C.3. Steady state The steady state is reached at a date Tm when the carbon budget is reached. After this date, emission are null in every sector and investment is used to counterbalance depreciation: ∀t ≥ Tm , ṁt = 0 =⇒ ∀i, ai,t = āi =⇒ xi,t = δi āi (C.5) Denoting Ti the date when the all emissions in sector i are capped (ai,t = āi ), it is immediate that Tm = maxi (Ti ). 25 Appendix C.4. Slackness conditions The complementary slackness conditions mean that the carbon price is constant before the steady state: ∀t < Tm , mt < B & φt = 0 =⇒ µt = µ (C.6) (C.7) and the social costs of the sectoral potentials are null before the respective dates Ti : ∀t < Ti , ai,t < āi & λi,t = 0 (C.8) Following the demonstration for the case of one single sector (Appendix B) yields: ∀(i, t), (r + δi ) ci 0 (xi,t ) − dci 0 (xi,t ) = µert − λi,t dt (C.9) Solving this differential equation for t ≤ Ti and using the boundary conditions xi,Ti = δi āi yields: ci 0 (x?i,t ) = µert Z Ti e−δi (θ−t) dθ + e−(δ+r)(Ti −t) c0i (δi āi ) (C.10) t Appendix C.5. The decarbonization dates Ti differ across sectors Let two sectors {1, 2} be such that they exhibit the same investment cost function, the same depreciation rate, but different abatement potentials: ∀x > 0, c01 (x) = c02 (x), δ1 = δ2 , ā1 > ā2 Suppose that the bigger sector does not take longer to decarbonize (i.e. T1 ≤ T2 ). Optimal MICs would then be lower in the first sector (Coroll. 3). This would lead to lower investment, hence lower abatement, in the first sector than in the second sector until the first sector decarbonizes (∀t ≤ T1 ), and to: ā1 = a1 (T1 ) ≤ a2 (T1 ) ≤ a2 (T2 ) = ā2 which would contradict the assumption that ā1 > ā2 . Therefore, a greater potential ā1 > ā2 has to lead to a further optimal decarbonizing dates T1 > T2 . Similarly, a sector where abatement capital is more expensive, in the sense that ∀x > 0, c01 (x) > c02 (x) (while ā1 = ā2 ) also takes longer to decarbonize (i.e. T1 > T2 ). Indeed, the same decarbonization dates (thus same marginal investment cost) would result in less investment, hence less abatement, in the more expensive sector. 26 Appendix D. An intuitive approach to the marginal implicit rental cost of capital Consider an investment strategy that increases abatement in a sector at one date while keeping the rest of the abatement trajectory unchanged (Fig. D.7). Such a strategy is similar to renting one unit of capital to reduce emissions during one time period. It reduces one unit of GHG, but leaves unchanged any opportunity to invest later in the same sector. Figure D.7: From a given investment pathway (xi,t ) leading to the abatement pathway (ai,t ), saving one more unit of GHG at a date θ without changing the rest of the abatement pathway, as in (ãt ), requires to invest one more unit at θ and (1 − δi dθ) less at θ + dθ, as (x̃t ) does. From an existing investment pathway (xi,t ) leading to an abatement pathway (ai,t ), the social planner may increase investment by one unit at time θ and immediately reduce investment by 1 − δ dθ at the next period θ + dθ. The resulting investment schedule (x̃t ) leads to an abatement pathway (ãt ) that abates one supplementary unit of GHG between θ and θ + dθ (Fig. D.7). Moving from (xi,t ) to (x̃t ) costs: 1 (1 − δi dθ) 0 P= ci (xi,θ+ dθ ) (D.1) ci 0 (xθ ) − dθ (1 + r dθ) For marginal time lapses, this tends to : P −−−−→ (r + δi ) ci 0 (xθ ) − dθ→0 dci 0 (xθ ) dθ (D.2) P tends to the cost of renting one unit of abatement capital at t. Appendix E. Proof of the expression of `t in Prop. 5 Let h be a marginal physical investment in abatement capital made at time t in sector i (expressed in tCO2 /yr per year). It generates an infinitesimal abatement flux that starts at h at time t and decreases exponentially at rate δi , leading to the total discounted abatement ∆A (expressed in tCO2 ): Z ∞ ∆A = er(θ−t) h e−δi (θ−t) dθ (E.1) θ=t = h r + δi (E.2) 27 This additional investment h brings current investment from xi,t to (xi,t + h). The additional cost ∆C (expressed in $) that it brings reads: ∆C = ci (xi,t + h) − ci (xi,t ) = h ci 0 (xi,t ) h→0 (E.3) The LAC `t is the division of the additional cost by the additional abatement it allows: ∆C ∆A `t = (r + δi ) ci 0 (xi,t ) `t = (E.4) (E.5) Appendix F. Levelized costs and implicit rental cost when investment costs are exogenous and linear Let It be the amount of investments made at exogenous unitary cost Qt to accumulate capital Kt that depreciates at rate δ: K̇t = It − δ Kt (F.1) Let F (Kt ) be a classical production function, and the price of output be the numeraire. Jorgenson (1967) defines current receipts Rt as the actual cash flow: Rt = F (Kt ) − Qt It he finds that the solution of the maximization program Z ∞ max e−rt Rt dt It (F.2) (F.3) 0 does not equalize the marginal productivity of capital to the investment costs Qt :24 FK (Kt? ) = (r + δ) Qt − Q̇t (F.4) He defines the implicit rental cost of capital Ct , as the accounting value: Ct = (r + δ) Qt − Q̇t (F.5) such that the solution of the maximization program is to equalize the marginal productivity of capital and the rental cost of capital: FK (Kt? ) = Ct (F.6) He finds that this is consistent with maximizing discounted economic profits, where the current profit is given by the accounting rule: Πt = F (Kt ) − Ct Kt (F.7) that is, profits equal revenue less the total cost of renting the capital. 24 Note that in our case, the marginal production of the abatement capital is simply one unit of abatement. 28 In this case, the (unitary) levelized cost of capital Lt is given by: Lt = (r + δ) Qt (F.8) And the levelized cost of capital matches the optimal rental cost of capital if and only if investment costs are constant: Q̇t = 0 ⇐⇒ FK (Kt? ) = Lt 29 (F.9)